考量隨機性之混合型流線式生產系統可靠度分析

Abstract

混合型流線式生產系統(Hybrid flow shop)為一種常見的製造系統，是一組作業依照指定順序、經過多個階段的生產流程，其中各階段可能擁有數台機器運作，以實現高效率生產。在混合型流線式生產系統的研究議題上，多數研究著重於探討理想狀態下的調度問題(Scheduling problem)，討論系統於特定目標的最佳化生產，忽略其中可能牽涉到的不確定性。本研究除了針對生產系統滿足產量需求的調度問題進行探討之外，考量生產系統所使用機器性能會隨時間衰退(Degradation)，而引入可靠度工程(Reliability engineering)裡的衰退概念，評估生產系統能夠滿足產量需求的可能性，並以機率呈現；本研究也考量每台機器單位時間產量的隨機性(Randomness)，重新探討前述問題。研究結果發現，源由於機器性能的衰退、可靠度的下降，調度問題分析所得的生產系統原可滿足產量需求，但隨著時間演進，滿足需求的機率逐漸降低；若再考量每台機器單位時間產量的隨機性，則生產系統滿足產量需求的機率將會降至更低。為求改善，本研究進一步探討機器與系統預防維護(Preventive maintenance)的成效，最後並依定期維護(Scheduled maintenance)和狀態維護(Condition based maintenance)兩種不同考量，提出三套不同維護策略，以優化生產系統，並供決策者參考。

The Hybrid Flow Shop (HFS) is a common manufacturing system where a sequence of operations is carried out across multiple stages in a predefined order. Each stage may contain multiple machines operating simultaneously to achieve high production efficiency. In research related to hybrid flow shops, most studies focus on the scheduling problem under ideal conditions, aiming to optimize specific objectives of the system while often neglecting potential uncertainties. This study not only investigates the scheduling problem to meet production demand but also considers machine performance degradation over time, incorporating the concept of degradation from reliability engineering to assess the probability that the production system can meet the required output. Furthermore, the study considers the randomness in each machine's unit-time output, and reexamines the problem from this perspective. The results indicate that, due to machine degradation and decreased reliability, a production system that was initially capable of meeting output demand may experience a decline in this probability over time. When randomness in machine productivity is also taken into account, the likelihood of meeting production requirements becomes even lower. To address this issue, the study further explores the effectiveness of preventive maintenance at both the machine and system levels. Based on two different maintenance approaches—scheduled maintenance and condition based maintenance—the study proposes three distinct maintenance strategies to optimize the production system and provide practical guidance for decision-makers.